Web Personalisation for Users Protection: A Multi-agent Method

Abstract

“Interracial Breeding Destroys The White Race...”, “Jews and Arabs should both be kicked out of the White West.” are examples of sentences that are easy to find on the Internet. What can be done to help those who want to protect themselves from such discourse? Put another way, how can we personalise web browsers to protect the users who desire so? This is the difficult subject raised in this article and to which we propose a multi-agent solution. This solution involves a multi-dimensional linguistic analysis of the content of the documents and the role of the multi-agent system is to combine these dimensions. The multi-agent model we propose combines a pyramidal coordination structure with agent associations, two notions which we introduce herein and of which we demonstrate the relevance.